US20060039502A1

US20060039502A1 - Phase correlated quadrature amplitude modulation

Info

Publication number: US20060039502A1
Application number: US11/157,055
Authority: US
Inventors: Daniel Mahgerefteh; Xueyan Zheng; Parviz Tayebati
Original assignee: Individual
Current assignee: Ii Vi Optical Systems Inc; Photop Technologies Inc; Finisar Corp; Marlow Industries Inc; M Cubed Technologies Inc; LightSmyth Technologies Inc; Optium Corp; Coadna Photonics Inc; Epiworks Inc; Kailight Photonics Inc; II VI Delaware Inc; II VI Optoelectronic Devices Inc; II VI Photonics US LLC; Coherent Corp
Priority date: 2002-11-06
Filing date: 2005-06-20
Publication date: 2006-02-23
Also published as: US7742542B2

Abstract

A method of transmitting a base band binary data contained in respective successive time cells, the data being in the form of a signal obtained by amplitude modulation and frequency modulation of a carrier wave with a 0 bit data having a 0 bit mean amplitude and a 0 bit mean frequency and a 0 bit duration and a 1 bit data having a 1 bit mean amplitude and a 1 bit mean frequency and a 1 bit duration the improvement wherein the amplitude and frequency are modulated with the same binary base band signal whereby to decrease the spectral width of the resulting modulated signal below the spectral width of the base band signal. A method for transmitting a base band digital data signal having a bit period and a first bandwidth, comprising providing a carrier wave having a carrier amplitude and a carrier frequency, and synchronously modulating both the carrier amplitude and carrier frequency with the base band digital signal, so as to reduce the bandwidth of the modulated carrier wave below the bandwidth of the base band signal. A method for transmitting a base band digital data signal having a bit period and a first bandwidth, comprising providing a carrier wave having a carrier phase and a carrier frequency, and synchronously modulating both the carrier phase and carrier frequency with the base band digital signal, so as to reduce the bandwidth of the modulated carrier wave below the bandwidth of the base band signal.

Description

REFERENCE TO PENDING PRIOR PATENT APPLICATIONS

This patent application:
(i) is a continuation-in-part of pending prior U.S. patent application Ser. No. 10/289,944, filed Nov. 6, 2002 by Daniel Mahgerefteh et al. for POWER SOURCE FOR A DISPERSION COMPENSATION FIBER OPTIC SYSTEM (Attorney's Docket No. TAYE-59474-00006);
(ii) is a continuation-in-part of pending prior U.S. patent application Ser. No. 10/308,522, filed Dec. 3, 2002 by Daniel Mahgerefteh et al. for HIGH-SPEED TRANSMISSION SYSTEM COMPRISING A COUPLED MULTI-CAVITY OPTICAL DISCRIMINATOR (Attorney's Docket No. TAYE-59474-00007);
(iii) is a continuation-in-part of pending prior U.S. patent application Ser. No. 10/680,607, filed Oct. 6, 2003 by Daniel Mahgerefteh et al. for FLAT DISPERSION FREQUENCY DISCRIMINATOR (FDFD) (Attorney's Docket No. TAYE-59474-00009);
(iv) claims benefit of pending prior U.S. Provisional Patent Application Ser. No. 60/581,076, filed Jun. 18, 2004 by Daniel Mahgerefteh et al. for PHASE CORRELATED AMPLITUDE MODULATION (Attorney's Docket No. TAYE-42 PROV); and
(v) claims benefit of pending prior U.S. Provisional Patent Application Ser. No. 60/615,834, filed Oct. 4, 2004 by Daniel Mahgerefteh et al. FOR PHASE CORRELATED QUADRATURE AMPLITUDE MODULATION (Attorney's Docket No. TAYE-45 PROV).
The five above-identified patent applications are hereby incorporated herein by reference.

DESCRIPTION OF THE INVENTION

Presently a favored method of data transmission for internet traffic from central office to the home is asymmetric digital subscriber line technique over twisted wires. The twisted wires used for telephone lines have a limited bandwidth and were designed for voice traffic, which requires <4 kHz bandwidth. Download of internet data requires higher bandwidth on the order of MHz. However, as new applications are found for internet, higher bandwidth traffic will be needed to avoid increased download time.
At present the rate of data transmission is limited by two factors: 1. Attenuation of high frequencies caused by large capacitance and resistance in the line. 2. Increased error rate caused by distortion resulting from the group delay dispersion of the line. In order to increase the amount of information that is transmitted it is important to choose a modulation format that produces a narrower bandwidth.
The spectrum of a digitally modulated signal is determined by the product of the spectrum of the modulated pulse and the spectrum of the random digital data. The resulting spectrum can therefore be narrowed by choosing the modulation format, which determines the spectrum of the information carrying signal, and by the coding scheme used to represent the random data. A number of modulation formats, such as frequency shift keying (FSK) and quadrature amplitude modulation (QAM), and pulsed amplitude modulation (PAM) each use a different scheme to carry the data with varying degrees of spectral efficiency as described in J. G. Prokais, “Digital Communication,” hereby incorporated by reference. Adding correlation to the random bit sequence, or coding in general, can also reduce the spectral bandwidth. Duobinary coding is one example of this approach.
In the present invention we disclose a method to decrease the spectral width of the digitally modulated signal by introducing correlation in the digital bits without coding. This method applies to all forms of digital communication regardless of the medium or the signal character. So it applies equally well to classical electrical communication though twisted pair wire, cable, as well as wireless communication, and optical communication. The only requirement is that there is a carrier wave, which is modulated.

PRIOR ART

A digitally modulated signal on a carrier wave can be represented mathematically as
E(t)=A(t)exp(−iω _c t+iφ(t)), (1)
where ω_cis the carrier frequency, A(t) is the amplitude envelop, and φ(t) is the time varying phase. For a conventional pulse amplitude modulation (PAM), the amplitude carries the digital modulation in form of pulses representing 1 and 0 bits, while for conventional phase shift keying (PSK) format the phase is modulated by the digital signal. The digital signal is in form of pulses, g(t), separated by the bit period, T and can be written as $\begin{matrix} d (t) = \sum_{n = - \infty}^{+ \infty} a_{n} g (t - nT) & (2) \end{matrix}$
where a_nare the digital bits.
In order to increase the symbol rate multidimensional formats are used in which both the amplitude and phase carry information. This can increase bandwidth efficiency somewhat but at the cost of a higher required signal to noise ratio at the receiver. A favorite scheme for reducing bandwidth is called duobinary modulation. In this case the digital bits are encoded by a delay add circuit according to b_n=a_n+a_n+1. By adding this correlation to the initially random, uncorrelated bit sequence a_n, the spectrum of the bit sequence b_n, is reduced in width by a factor of 2. However, the resulting digital signal has 3 levels, and requires a decoder at the receiver to interpret the bits. In optical duobinary the decoding is performed at the transmitter by using a Mach-Zhender modulator. The encoder and decoder are added complexity and are not desirable.

DESCRIPTION OF THE INVENTION

It is an objective of the present invention to reduce the bandwidth of a digital signal by encoding the same digital information in the amplitude and frequency of the carrier wave. This redundant modulation introduces correlation into the digital signal and reduces the spectral width without the need for a coder/decoder pair. More generally for a multidimensional digital signal, where amplitude, phase and frequency of the carrier may carry digital information, the same information is encoded in more than one dimension, such as in amplitude and phase, multi-level amplitude and multi-level frequency for the same effect, reducing bandwidth of resulting signal.
In the preferred embodiment of the present invention the signal is amplitude and frequency modulated simultaneously with the same base band digital data, we call this phase correlated amplitude modulation (PCAM):
E(t)=A(t)exp[−iω _c t−i2πf _d∫_−∞ A(t′)/A ₀ dt′], (3)
where f_dis the peak to peak frequency deviation, and A₀is the absolute value of the peak amplitude of the envelope signal. Here it is assumed that the envelope A(t) has a zero average. Otherwise the average is subtracted in the integral of Eq. 3. The carrier frequency is therefore modulated and is given by: $\begin{matrix} f (t) = - \frac{1}{2 π} \frac{ⅆ ϕ (t)}{ⅆ t} = f_{d} A (t) / A_{0} . & (4) \end{matrix}$
FIG. 1 shows an example of a PCAM signal for a portion of a binary digital signal. Note that the amplitude and frequency code for the same bit sequence, creating a correlation between adjacent bits through the phase of the signal. In this example the frequency excursion is chosen to induce a π phase shift for period T of the bit period: f_dT=½. In this case 1 bits have the same phase if they have even number of zeros or no zeros between them and are π out of phase if they have an odd number of 0 bits between them. It is an embodiment of the present invention that this phase correlation reduces the spectral width of the digital signal.
The power spectrum of a digital signal carrying signal in Eq. 2 is given in Proakis, “Digital Communications,” to be $\begin{matrix} S (f) = \frac{1}{T} {\langle G (f) \rangle}^{2} Φ_{ii} (f), & (5) \end{matrix}$
where G(f) is the Fourier transform of the pulse shape g(t), and Φ_ii(f) is the power density spectrum of the information sequence {a_n}, which is defined in terms of the autocorrelation function of the bit sequence as $\begin{matrix} Φ_{ii} (f) = \sum_{m = - \infty}^{\infty} ϕ_{ii} (m) \exp (- ⅈ2π fmT), where & (6) \\ ϕ_{ii} (m) = \frac{1}{2} 〈 a_{n} a_{n + m}^{*} 〉 & (7) \end{matrix}$
is the expectation value of the autocorrelation of the bit sequence. Consider the example in which the frequency excursion produces a π phase shift over a bit period:
f _d T=1/2 (8)
In this case the digital signal in Eq. (1) can be written in terms of a set of coefficients, b_n=1, 0, −1 which are not statistically independent. However, the probability of occurrence of a bit sequence in b_ncan be derived from the probability of occurrence of the parent sequence a_nfrom which it was mapped. For example a 1 0 sequence in the original uncorrelated pattern a_n, which has a probability ¼, maps to either a 1 0 or a −1 0 with equal probability. Hence the probability of getting a 1 0 or a −1 0 in the new correlated sequence b_nin the case is Using FIG. 1 as a reminder of the rule imposed by the frequency modulation, the neighboring b_nsequences re related such:
If b_n=1, then b_n±1={0,1} with probability {¼,¼} for each case,
If b_n=0, then b_n±1={−1,0,1} with probabilities {¼,½,¼}
If b_n=−1, then b_n±1={0,−1} with probabilities {¼,¼} (9)
For neighboring bits farther than one bit away, b_n±m, m>=2, it can be shown that the correlation sums to zero. So the correlation bit length for this example is 1 and the bits are uncorrelated farther than 1 bit away. Using the definition of expectation value, in Eq. (7), and the relations in Eq. (9) the autocorrelation of the sequence b_nis given by $\begin{matrix} ϕ_{ii} (m) = \lim_{N -> \infty} \frac{1}{2 N} \sum_{n = - N}^{+ N} b_{n} b_{m + n}^{*} = {\begin{matrix} \frac{1}{2} & m = 0 \\ \frac{1}{4} & m = \pm 1 \\ 0 & \langle m \rangle \geq 2 \end{matrix} & (10) \end{matrix}$
Substituting Eq. (10) for the autocorrelation of the bit sequence into Eq. (6), the power density spectrum of the bit sequence, {b_n} is given by the first two terms to be
Φ_ii(f)=4 cos²(πfT), (11)
which has zeros at f=±1/2T, half the bit rate of the digital data. Hence the Phase correlated amplitude modulated digital signal has a spectral width that is ½ that of a standard amplitude modulated signal. Hence the PCAM produces a narrow spectrum similar to duobinary coded digital signal but without the coding.
FIG. 2 shows the spectra of a digital signal represented by pulse amplitude modulation (PAM), frequency modulation (FSK), and phase correlated amplitude modulation (PCAM). It is clear that the PCAM spectrum is narrower than both FSK and PAM. Note that in addition to producing a narrower spectrum the CPAM format also removes the carrier. In this example, where f_dT=1/2, half of the 1 bits have zero phase while the other half have π phase. Hence the integral of the signal over time, which determines the dc component of the power spectrum, averages to zero; there is no carrier.
The reduction in spectral width is related to what we call a phase correlation length, and is a function of the frequency excursion f_d. For the example f_dT=1/2, the correlation length is 1 bit long. (According to Eq. 10 the correlation is 0 for m>1). For f_dT=1, all pulses are in phase and there is no phase correlation. Hence the spectrum is not narrowed compared to the AM case. For f_dT<1/2 there is some spectral narrowing compared to AM case, however, and the carrier is reduced as shown in FIG. 3. In this Figure the carrier for the AM case has been artificially removed to make comparison easier.
The present invention can be generalized to multidimensional modulation, in which case one or more dimensions, such as amplitude, phase, or frequency carry the same data to create correlation in the modulation.
In another embodiment of the present invention, the phase correlated amplitude concept is applied to quadrature amplitude modulation (QAM). In prior art a QAM signal is mathematically represented by
E(t)=f(t)cos(2πf _c t)+g(t)sin(2πf _c t) (12)
where f(t) and g(t) represent two distinct digital signals each carrying independent random digital bit sequence, as given in Eq. 2. According to the present invention a phase correlated QAM signal is obtained by modulating the carrier frequency of the sin and cos components by the digital data according to:
Q(t)=f(t)cos [2πf _c t+2πf _d∫_−∞ f(t′)/f ₀ dt′]+g(t)sin [2πf _c t+2πf _d∫_−∞ g(t′)/g ₀ dt′] (13)
In this way the carrier frequency for the two quadrature signals is modulated and is given by
f _sc(t)=f _c +f(t′)/f ₀
f _cc(t)=f _c +g(t′)/g ₀ (14)
where f_scis the carrier frequency for the sine component and f_ccis the carrier frequency for the cosine component. The frequency excursion f_dis chosen to be related to the bit period. As an example, when f_dT=1/2, there will be a π phase shift between 1 bits having an odd number of zeros in between them. The correlation between the bits created by the synchronous FM/AM modulation reduces the bandwidth in a similar manner to that shown above.
In another embodiment the simultaneous frequency and amplitude modulation is applied to a multilevel QAM signal. In this case, two independent bit sequences having values 0,1 in one case, and 0,2 in the other are added to generate a three level signal. The frequency modulation of the first signal, having 1 as the high level is modulated according to the amplitude signal where f_dT=1/2. For the second signal, where the high level is represented by amplitude 2, the synchronous frequency excursion, which also follows the amplitude is set to f_dT=3/2. In general this signal can have a cosine carrier and a second such signal with independent data can have a sine carrier. The addition of these two quadrature signals produces a 4 level QAM. This prescription for signal generation can be generalized to multilevel QAM by associating a different excursion frequency to each distinct high level such that f_dT=(2k−1)/2, and k=1,2,3, . . . .

Claims

1. A method of transmitting a base band binary data contained in respective successive time cells, the data being in the form of a signal obtained by amplitude modulation and frequency modulation of a carrier wave

with a 0 bit data having a 0 bit mean amplitude and a 0 bit mean frequency and a 0 bit duration and a 1 bit data having a 1 bit mean amplitude and a 1 bit mean frequency and a 1 bit duration

the improvement wherein

the amplitude and frequency are modulated with the same binary base band signal whereby to decrease the spectral width of the resulting modulated signal below the spectral width of the base band signal.

2. A method as in claim 1

wherein the product of the 0 bit time duration and the difference between the 1 bit frequency and the 0 bit frequency is substantially equal to ½.

3. A method for transmitting a base band digital data signal having a bit period and a first bandwidth, comprising providing a carrier wave having a carrier amplitude and a carrier frequency, and synchronously modulating both the carrier amplitude and carrier frequency with the base band digital signal, so as to reduce the bandwidth of the modulated carrier wave below the bandwidth of the base band signal.

4. A method as in claim 3, wherein the product of the frequency excursion of the modulated carrier wave and the bit period of the base band digital signal is substantially equal to an odd integer multiple of ½.

5. A method as in claim 3, wherein the amplitude excursion of the modulated carrier wave is higher than 10 decibels.

6. A method for transmitting a base band digital data signal having a bit period and a first bandwidth, comprising providing a carrier wave having a carrier phase and a carrier frequency, and synchronously modulating both the carrier phase and carrier frequency with the base band digital signal, so as to reduce the bandwidth of the modulated carrier wave below the bandwidth of the base band signal.